Bursting with Potential: How Sensorimotor Beta Bursts Develop from Infancy to Adulthood

Abstract

Beta activity is thought to play a critical role in sensorimotor processes. However, little is known about how activity in this frequency band develops. Here, we investigated the developmental trajectory of sensorimotor beta activity from infancy to adulthood. We recorded EEG from 9-month-old, 12-month-old, and adult humans (male and female) while they observed and executed grasping movements. We analyzed "beta burst" activity using a novel method that combines time-frequency decomposition and principal component analysis. We then examined the changes in burst rate and waveform motifs along the selected principal components. Our results reveal systematic changes in beta activity during action execution across development. We found a decrease in beta burst rate during movement execution in all age groups, with the greatest decrease observed in adults. Additionally, we identified three principal components that defined waveform motifs that systematically changed throughout the trial. We found that bursts with waveform shapes closer to the median waveform were not rate-modulated, whereas those with waveform shapes further from the median were differentially rate-modulated. Interestingly, the decrease in the rate of certain burst motifs occurred earlier during movement and was more lateralized in adults than in infants, suggesting that the rate modulation of specific types of beta bursts becomes increasingly refined with age.SIGNIFICANCE STATEMENT We demonstrate that, like in adults, sensorimotor beta activity in infants during reaching and grasping movements occurs in bursts, not oscillations like thought traditionally. Furthermore, different beta waveform shapes were differentially modulated with age, including more lateralization in adults. Aberrant beta activity characterizes various developmental disorders and motor difficulties linked to early brain injury, so looking at burst waveform shape could provide more sensitivity for early identification and treatment of affected individuals before any behavioral symptoms emerge. More generally, comparison of beta burst activity in typical versus atypical motor development will also be instrumental in teasing apart the mechanistic functional roles of different types of beta bursts.

Keywords: EEG; action observation; beta burst; infant reaching and grasping; sensorimotor cortex.

Figure 1.

Beta peak frequency increases with age, and beta activity consists of transient burst events in infancy and adulthood. a, Periodic power spectral density in the combined C3 and C4 cluster of the 9-month-old participants. Dark lines indicate the mean over participants. Shaded areas represent SE. Gray shaded region represents the limits of the identified beta band. Inset, The electrodes included in the analysis. b, Mean lagged coherence in the combined C3 and C4 cluster across all 9-month-old participants for a range of frequencies and lags. There is high lagged coherence in the alpha band over a wide range of lags, but beta lagged coherence rapidly decreases after 2 cycles. c, Mean lagged coherence in the combined C3 and C4 cluster across all 9-month-old participants for lags of 2, 3, and 4 cycles. Solid lines indicate the mean. Shaded areas represent SE. A prominent peak appears in the α range at 2-4 cycles, while a beta peak is visible only at 2 cycles. d-f, Same as in a-c, for 12-month-old participants. g-i, Same as in a-c, for adult participants.

Figure 2.

Power and lagged coherence localize beta to the C3 and C4 clusters. a, Topography of beta band periodic power (after subtraction of the aperiodic spectral density), averaged over 9-month-old participants. White circles represent electrodes included in the C3 and C4 clusters. Power in the beta band is most prominent in peripheral electrodes. b, Beta lagged coherence topographies at (from left to right) 2, 3, and 4 cycles averaged over 9-month-old participants. Lagged coherence in the beta band localizes to the C3 and C4 electrodes and decreases rapidly after two cycles. c, d, Same as in a, b, for the 12-month-old participants. e, f, Same as in a, b, for the adult participants.

Figure 3.

a-f, Alpha/μ occurs as a sustained oscillation in the C3 and C4 clusters. Same as in Figure 2, for the α/μ band. Power in the α/μ band is strongest in peripheral, central, and occipital electrodes in infants, and in central and frontal electrodes in adults. Lagged coherence localizes α/μ to the C3 and C4 electrodes and remains high up to at least 4 cycles at all ages.

Figure 4.

Beta burst timing is not a pure Poisson process. a, The distribution of the coefficient of variation of the IBIs over participants from each age group (dark shaded histograms), and from surrogate data with random burst times (light shaded histograms). b, The mean IBI distribution for each age group (solid lines; shaded area represents the SE).

Figure 5.

Beta bursts are diverse in infancy and adulthood. a-d, The distributions of beta burst peak amplitude (a), peak frequency (b), frequency span (c), and duration (d) for the 9-month-old, 12-month-old, and adult participants over conditions, epochs, and clusters. e-h, The distributions of burst peak amplitude (e), peak frequency (f), frequency span (g), and duration (h) for the (rows, from top to bottom), 9-month-old, 12-month-old, and adult participants for the ipsilateral and contralateral electrode clusters. i, The mean burst rate (dashed lines; shaded area represents SE) and mean beta amplitude (solid lines; shaded area represents SE) in the ipsilateral and contralateral clusters for the 9-month-old participants in the execution condition. Colored dots and asterisks represent times in which beta amplitude or burst rate significantly deviated from baseline. j, Same as in i, for the 12-month-old participants. k, Same as in i, for the adult participants.

Figure 6.

Infant and adult bursts have qualitatively similar burst waveforms. a, The median waveform (thick red line) over all detected beta bursts from 9-month-old participants had a wavelet-like shape, but there was great variability in the waveforms of individual bursts (thin colored lines). b, Same as in a, for the 12-month-old participants. c, Same as in a, for the adult participants. d, Correspondence between time points (dashed lines) of 9-month-old beta burst waveforms (red) and adult beta burst waveforms (green). Inset, The alignment curve (solid line) resulting from dynamic time warping of beta burst waveforms from 9-month-old participants to those of adults. The dashed line indicates the alignment curve for two already aligned signals. e, Same as in d, for the 12-month-old participants. f, The adult beta burst waveforms were, on average, shorter in absolute duration and smaller in amplitude than the mean infant bursts. g, Normalization and dynamic time warping revealed that the infant and adult beta bursts have qualitatively similar waveform shapes.

Figure 7.

Global versus age group-specific PCA. a, Absolute value of the correlation coefficient between eigenvectors from the PCA ran only on 9-month-old bursts, and those from the global PCA. b, c, Same as in a, for 12-month-olds (b) and adults (c). d, For each burst detected in 9-month-olds, the score for each PC from PCA ran on only the 9-month-old bursts (x axis) versus the score from the PCA ran on all bursts over all ages. e, f, Same as in d, for 12-month-olds (e) and adult participants (f). g, Absolute value of Pearson's correlation coefficient between scores for 9-month-old infant bursts from PCA applied only to 9-month-old bursts and the global PCA, for each PC. h, i, Same as in g, for 12-month-olds (h) and adult participants (i).

Figure 8.

PCA on warped versus unwarped infant burst waveforms. a, For each burst detected in 9-month-olds, the score for each PC from PCA ran on only the warped 9-month-old bursts (x axis) versus the score from the PCA ran on only the unwarped 9-month-old bursts. b, Same as in a, for 12-month-olds. c, Absolute value of Pearson's correlation coefficient between scores for 9-month-old infant bursts from PCA applied only to warped 9-month-old bursts and the PCA applied to unwarped 9-month-old bursts, for each PC. d, Same as in c, for 12-month-olds.

Figure 9.

Contralateral burst waveform motifs are increasingly rate-modulated during movement from infancy to adulthood. a, The mean normalized and warped waveforms of beta bursts with scores in four quartiles of PC 3 scores (colored lines) and the mean overall burst waveform (black). b, The mean baseline-corrected rate of bursts with scores in each PC 3 score quartile (colored lines; shaded area represents SEM) over the course of the (columns, from left to right) trial start, first touch, grasp completion, and trial end epochs in the contralateral center cluster for 9-month-old (top row), 12-month-old (middle row), and adult (bottom row) participants. Colored dots represent where the burst rate in the corresponding score quartile is different from baseline. c, d, Same as in a, b, for PC 4. e, f, Same as in a, b, for PC 6.

Figure 10.

Ipsilateral beta burst motifs rate-modulated differently from contralateral motifs during movement in adulthood, but not infancy. a, The mean normalized and warped waveforms of beta bursts in with scores in four quartiles of PC 3 scores (colored lines) and the mean overall burst waveform (black). b, The mean baseline-corrected rate of bursts with scores in each PC 3 score quartile (colored lines; shaded area represents SEM) over the course of the (columns, from left to right) trial start, first touch, grasp completion, and trial end epochs in the ipsilateral central cluster for 9-month-old (top row), 12-month-old (middle row), and adult (bottom row) participants. Colored dots represent where the burst rate in the corresponding score quartile is different from baseline. c, d, Same as in a, b, for PC 4. e, f, Same as in a, b, for PC 6.